Abstract If fractional crystallization coupled with some crustal assimilation (AFC) is the dominant process for generating evolved magmas in the Earth's crust, large quantities of igneous cumulates must be present in the rock record. However, despite some reported occurrences, intermediate to silicic rocks characterized by an unambiguous crystal accumulation signature have remained elusive. The signature is expected to be subtle in viscous silicic magmas, and sensitive criteria need to be developed to test whether these cumulates are truly rare or simply difficult to discern. We propose models that use specific trace element concentrations and ratios (Zr, Ba, Eu/Eu*, and Zr/Hf) within a well-constrained framework for the physical extraction of liquid from crystals (mostly occurring between 50 and 70 vol.% crystals) to identify silicic cumulates. The results of these models indicate that crystal accumulation (or melt extraction) has occurred in multiple natural examples around the world (both in plutonic and volcanic realms). However, these examples were selected because they represent conditions that are most favorable (high amount of melt extraction, well-constrained initial liquid compositions) for unambiguously identifying cumulate residue; in many cases, we show that the trace element signature for crystal accumulation will remain largely concealed. In addition, the different liquid lines of descent related to the various tectonic settings on Earth (dry-reduced in hot spots-divergent margins and wetter-more oxidizing in arc environments) will lead to different trace element evolution and each criterion must be applied on a case-by-case basis. Despite these limitations, the undeniable but subtle presence of silicic cumulates in the mid- to upper-crust lends support to the idea that igneous differentiation is dominantly driven by melt extraction from partially crystalline reservoirs.